Periodic Reporting for period 1 - Dia-Pol (Dia-Pol: Polarization or dialogue? A deep learning study of the Black lives matter and Me Too online social movements)
Reporting period: 2021-07-01 to 2023-06-30
This study explored whether interactions between social movements yield polarization or encourage dialogue using the cases of the Black Lives Matter (BLM) and Me Too (MT) movements. The project had 3 objectives: (a) pioneer a novel research agenda on the dynamics of movement interactions, filling a void in political science, sociology, communication, and race and gender studies; (b) devise an algorithm to aid stakeholders, e.g. citizens, activists and NGOs, in identifying divisive issues and crafting messages to navigate sensitivities; and (c) advance computational social science (CSS) within political science and at the host university by establishing a CSS lab at Bogazici.
Studying how social movements interact with each other and with their countermovements is crucial for understanding and mitigating polarization, especially on issues like race and gender since even basic demands for equality can be misconstrued and weaponized by extremist groups. Recent events, e.g. the Kyle Rittenhouse trial, demonstrate how right-wing groups can distort BLM's messages. Understanding how these distortions occur is essential for developing strategies to counter them. By analyzing social movement interactions, we can identify how disagreements escalate, how misinformation spreads, and how echo chambers amplify extreme views. Understanding these processes helps develop more effective strategies for de-escalating tensions and fostering dialogue. In a polarized environment, NGOs and activists must frame messages clearly and concisely, while being mindful of how their words might be misinterpreted. Dia-Pol’s algorithm helps identify the key themes and concerns in public discourse. This information, now available on the Dia-Pol platform, can be used to strategically craft messages. Also, political science, sociology, communication, and data science play important roles in understanding and mitigating polarization. To foster interdisciplinary collaboration and promote CSS, this project organized workshops and training. This project utilized open science practices (GitLab) for data analysis. It developed a novel theory to study variations within countermovements, examined the effect of BLM protests on state-level policing reforms and court decisions, and studied the media coverage of BLM protests.
Studying how social movements interact with each other and with their countermovements is crucial for understanding and mitigating polarization, especially on issues like race and gender since even basic demands for equality can be misconstrued and weaponized by extremist groups. Recent events, e.g. the Kyle Rittenhouse trial, demonstrate how right-wing groups can distort BLM's messages. Understanding how these distortions occur is essential for developing strategies to counter them. By analyzing social movement interactions, we can identify how disagreements escalate, how misinformation spreads, and how echo chambers amplify extreme views. Understanding these processes helps develop more effective strategies for de-escalating tensions and fostering dialogue. In a polarized environment, NGOs and activists must frame messages clearly and concisely, while being mindful of how their words might be misinterpreted. Dia-Pol’s algorithm helps identify the key themes and concerns in public discourse. This information, now available on the Dia-Pol platform, can be used to strategically craft messages. Also, political science, sociology, communication, and data science play important roles in understanding and mitigating polarization. To foster interdisciplinary collaboration and promote CSS, this project organized workshops and training. This project utilized open science practices (GitLab) for data analysis. It developed a novel theory to study variations within countermovements, examined the effect of BLM protests on state-level policing reforms and court decisions, and studied the media coverage of BLM protests.
This project achieved three key goals: pioneering research on movement interactions and polarization, developing an algorithm for identifying divisive issues, and advancing computational social science (CSS) in political science.
To address Objective 1, I researched the intricate dynamics of movement interactions and polarization. By compiling and analyzing two big Twitter datasets encompassing the #BlackLivesMatter and #MeToo movements, I gained insights into how online discussions unfold around crucial topics, revealing issues of support, opposition, and interconnectivity. This work, outlined in 6 co-authored papers (one published, five under review) and presented at academic conferences and online talks, sheds new light on polarization and social justice in democracies.
Concerning Objective 2, I developed an algorithm to analyze social media data and identify the key themes that structure online debates around divisive issues. This algorithm was incorporated into the Dia-Pol website (https://research.diapols.com/(opens in new window)) to provide open access to these insights. This user-friendly platform, featuring dynamic visualizations and comparative analysis tools, has become a valuable resource for researchers and the public alike, fostering transparency and raising awareness about critical social movements. To the best of my knowledge, it is the first website that summarizes the main topics of MT and BLM.
Regarding objective 3, I dedicated significant efforts to advancing CSS education and research. Through a series of international training sessions and workshops, including those at the American Political Science Association and the Summer Institutes in Computational Social Science, I trained about 122 researchers with practical skills in machine learning, natural language processing, and network analysis. In Turkey, I offered training at Tevitol, a high school for gifted students on Jun. 12, 2022. At Bogazici, I established a CSS lab and trained 55 students in machine learning, natural language processing, and network analysis. Our students have secured prestigious doctoral positions at top universities like Northwestern, Stanford, and Oxford, while others have embarked on successful careers in data science. Moreover, I offered two data science courses at the political science department: POLS 204 Advanced Statistics and POLS 598 Research & Reading in analyzing social media data. About 44 students took these classes. Three students wrote dissertations using CSS methods, while three others landed prestigious doctoral positions at Northwestern, Stanford, and Oxford thanks to their CSS skills. Relatedly, I published a book chapter on incorporating computational social science in Political science in an edited book entitled Opportunities and Challenges for Computational Social Science Methods (https://orcid.org/0000-0001-5479-4801(opens in new window)).
To address Objective 1, I researched the intricate dynamics of movement interactions and polarization. By compiling and analyzing two big Twitter datasets encompassing the #BlackLivesMatter and #MeToo movements, I gained insights into how online discussions unfold around crucial topics, revealing issues of support, opposition, and interconnectivity. This work, outlined in 6 co-authored papers (one published, five under review) and presented at academic conferences and online talks, sheds new light on polarization and social justice in democracies.
Concerning Objective 2, I developed an algorithm to analyze social media data and identify the key themes that structure online debates around divisive issues. This algorithm was incorporated into the Dia-Pol website (https://research.diapols.com/(opens in new window)) to provide open access to these insights. This user-friendly platform, featuring dynamic visualizations and comparative analysis tools, has become a valuable resource for researchers and the public alike, fostering transparency and raising awareness about critical social movements. To the best of my knowledge, it is the first website that summarizes the main topics of MT and BLM.
Regarding objective 3, I dedicated significant efforts to advancing CSS education and research. Through a series of international training sessions and workshops, including those at the American Political Science Association and the Summer Institutes in Computational Social Science, I trained about 122 researchers with practical skills in machine learning, natural language processing, and network analysis. In Turkey, I offered training at Tevitol, a high school for gifted students on Jun. 12, 2022. At Bogazici, I established a CSS lab and trained 55 students in machine learning, natural language processing, and network analysis. Our students have secured prestigious doctoral positions at top universities like Northwestern, Stanford, and Oxford, while others have embarked on successful careers in data science. Moreover, I offered two data science courses at the political science department: POLS 204 Advanced Statistics and POLS 598 Research & Reading in analyzing social media data. About 44 students took these classes. Three students wrote dissertations using CSS methods, while three others landed prestigious doctoral positions at Northwestern, Stanford, and Oxford thanks to their CSS skills. Relatedly, I published a book chapter on incorporating computational social science in Political science in an edited book entitled Opportunities and Challenges for Computational Social Science Methods (https://orcid.org/0000-0001-5479-4801(opens in new window)).
The research carried out within Dia-Pol empowers activists, human rights NGOs, and policymakers by providing evidence of the effectiveness of three strategies.
1. Activists are better off organizing small frequent demonstrations rather than large infrequent ones. Direct engagement with state-level policymakers is more cost-effective and yields greater results than lobbying at the federal level. Partisan divides at the federal level, often seen as an insurmountable barrier, do not necessarily hinder reform at the state level. 2. Activists are more likely to receive policy responses and public support for their cause if they organize smaller peaceful but frequent protests than larger violent protests. Online activism plays a crucial role in keeping issues at the forefront of the agenda. Online activists should strategically time when to emphasize and de-emphasize the particulars of a case about a victim. 3. Activists should use simple frames that are accessible to everyone. Simple frames are powerful in enlarging the support basis. However, these simple frames may fall short of addressing the grievances of intersectional groups. In this case, activists should strike a balance between simplicity and inclusivity for broad appeal and targeted messaging for specific issues. Activists should remain adaptable, considering the evolving dynamics of their movements and the diverse needs of their audience.
In this process, initiatives like Dia-Pol’s algorithm play a crucial role because they inform activists and NGOs about the sensitivities of the public, i.e. what issues the public sees as being intertwined to pre-empt misunderstandings and use the appropriate emphasis to enlarge the support basis. Such websites contribute to the democratization of research findings. While I am not the first to provide a user-friendly interface to visualize data, to the best of my knowledge, it is the first website that summarizes the main topics that shape the Twitter debates on BLM and MT. By democratizing research through the Dia-Pol platform, the project empowers a more informed and engaged citizenry. This has the potential to drive positive socio-economic changes by fostering public awareness of critical issues and equipping communities to advocate for themselves.
1. Activists are better off organizing small frequent demonstrations rather than large infrequent ones. Direct engagement with state-level policymakers is more cost-effective and yields greater results than lobbying at the federal level. Partisan divides at the federal level, often seen as an insurmountable barrier, do not necessarily hinder reform at the state level. 2. Activists are more likely to receive policy responses and public support for their cause if they organize smaller peaceful but frequent protests than larger violent protests. Online activism plays a crucial role in keeping issues at the forefront of the agenda. Online activists should strategically time when to emphasize and de-emphasize the particulars of a case about a victim. 3. Activists should use simple frames that are accessible to everyone. Simple frames are powerful in enlarging the support basis. However, these simple frames may fall short of addressing the grievances of intersectional groups. In this case, activists should strike a balance between simplicity and inclusivity for broad appeal and targeted messaging for specific issues. Activists should remain adaptable, considering the evolving dynamics of their movements and the diverse needs of their audience.
In this process, initiatives like Dia-Pol’s algorithm play a crucial role because they inform activists and NGOs about the sensitivities of the public, i.e. what issues the public sees as being intertwined to pre-empt misunderstandings and use the appropriate emphasis to enlarge the support basis. Such websites contribute to the democratization of research findings. While I am not the first to provide a user-friendly interface to visualize data, to the best of my knowledge, it is the first website that summarizes the main topics that shape the Twitter debates on BLM and MT. By democratizing research through the Dia-Pol platform, the project empowers a more informed and engaged citizenry. This has the potential to drive positive socio-economic changes by fostering public awareness of critical issues and equipping communities to advocate for themselves.